Comparison of on- and off-axis photorefraction with cycloplegic retinoscopy in infants

We have compared the performance of an off-axis (knife-edge) photorefractor with that of an on-axis (isotropic) system. Normal infants and children between the ages of 8 and 208 weeks were photographed with each camera both with and without cycloplegia. Refractive errors were estimated for each technique based on equations derived from ray-tracing. These refractions were compared to the results of retinoscopy under cycloplegia. Sensitivity and specificity of the two photorefraction systems were evaluated as a function of the magnitude of meridional hyperopia defined by retinoscopy. We also examined the effect of varying the photorefraction screening criterion. Thirteen percent of the infants in the screening sample presented with +3.50 diopters or more of meridional hyperopia. Using this level of ametropia as a referral criterion, the sensitivity and specificity of the off-axis system for infants without cycloplegia were 83% and 72%, respectively. For the on-axis system, sensitivity and specificity values were 85% and 53%. The use of cycloplegics did not significantly improve the performance of either system, but rather their use degraded the specificity of the on-axis system in the presence of moderate refractive errors. The results of the present study indicate that both on- and off-axis systems are effective in identifying highly ametropic infants, but that the off-axis system results in significantly fewer false positives. Moreover, the off-axis system has the advantages of an inherently greater dynamic range for a fixed camera design, and also more easily interpreted photographs.